testPar: Test Parameters For Peak Detection

Description Usage Arguments Details Value Note Author(s) See Also

Description

This function presents plots for evaluating the quality of the parameters used by xcmsSet from xcms.

Usage

1
testPar(xcmsRaw = file.choose(), ppm = 25, peakwidth = c(5, 25), snthresh = 5, prefilter = c(3, 100))

Arguments

xcmsRaw

An .mzXML file representing the data the parameters should be used for.

ppm

The sensitivity in the mz direction.

peakwidth

A vector with minumum and maximum peak width in seconds.

snthresh

The signal to noise threshold.

prefilter

Prefiltering parameters.

Details

This function runs xcmsRaw and findPeaks.centWave and returns a 5*5 grid plot of 25 randomly picked peaks from the run overlayed with their signal to noise. Additionally it creates .png files with plots of chromatograms for each 100 mz slice of the dataset with detected peak location. The .png files are writen to the working directory.

ppm
The ‘ppm’ parameter controls the width of the window used for finding regions of interest in the mz data. This parameter is not linked to the manufacturers ppm claims and should usually be quite a bit higher (around 25 for an LC-qTOF).

snthresh
The signal to noise is calculated by (maxo - baseline)/sd, where maxo is the maximum intensity of the peak, baseline is the mean of the data without the 10 % highest and lowest data points and sd is the standard deviation of the data without the 10 % highest and lowest data points. This value should be kept low (i.e. 5) as it is not very robust and exclusion of real peaks are likely to occur at higher values.

prefilter
The prefilter is a vector c(k, I) giving the number of peaks (k) with intensity above I that a given region of interest should contain for it to be kept for further analysis.

Value

Creates .png files in the working directory of name: mzXXX-YYY.png with XXX and YYY being the mz window of the plot.

Additionally a 5*5 grid plot is produced showing 25 randomly picked peaks and a 5*5 matrix giving the index of the peaks is written.

Note

A high number of false positives is unavoidable using current peak detection methods.

Author(s)

Thomas Dybdal Pedersen thomasp85@gmail.com

See Also

xcmsSet, findPeaks.centWave


thomasp85/pepmaps documentation built on May 31, 2019, 11:15 a.m.